Economics 448: Measures of Inequality 6 March 2014
1 2 The context Economic inequality: Preliminary observations 3
Inequality Economic growth affects the level of income, wealth, well being. Also want to study the distribution of income, wealth, well being. Two dimensions: efficiency for a given level of output, product at lowest cost. equity distribution who receives what? Winners and losers.
Equity: Social Justice In econ courses we spend probably 80 90% of time thinking (characterizing) economic efficiency. (Positive Economics; what is) Equity involves notions of justice what is a fair distribution of resources? (Normative economics; what should be) Pareto Optimality Criterion: no one can be made better off without some else being made worse off. Competitive equilibrium yields a Pareto outcome. But every point on the production possibility curve can support a competitive equilibrium. Pareto optimality is a weak criterion need additional tools to
Two motivations to study equity Social Justice philosophy. Rawls: veil of ignorance. Functional level how the distribution of resources affects economic growth. In Chapter 6 only study the measurement of inequality. In chapter 7 studies functional level; consequences of inequality.
The context Economic inequality: Preliminary observations Context Inequality is the fundamental disparity that permits one individual certain material choices, while denying another individual those very same choices. Will not spend much time on notions of social justice (what is fair?). Sen: Commodities and Capabilities. Oxford University Press. 1985.
The context Economic inequality: Preliminary observations Preliminary Observations Previously mentioned: Stocks versus Flows. Assets the accumulation of past income flows. Economic Mobility important. Can measure dispersion of income at point in time. But is there any churning in the distribution. Musical chairs? or Make like a statue?
The context Economic inequality: Preliminary observations Functional and Personal Distribution of Income Functional income distribution the returns to different factors of production (labor, capital, land). Need to know who owns the different factors. Would like to know for each individual income from labor, profits from capital, land rents. Understanding sources may affect how we judge the outcome. Functional distribution income informative on features of development.
The Challenge How to compare alternative distributions of income across many people? Pareto criterion gives us a partial ordering. May be able to rank order some distributions. But that criterion does not offer much guidance. Will develop inequality indexes that collapse the distribution (income, assets, etc.) into a single number. Scalar measures easy to report and understand. And offer a complete ordering.
Two Approaches Descriptive Use measures that describe inequality, without which make no direct use of Social Welfare. Normative Use measures of inequality directly connected to notion of Social Welfare and loss incurred by inequality Distinction between the approaches is sometimes blurred. We focus on Descriptive Measures, but take a quick tangent to consider the notion of Social Welfare.
Social Welfare A normative notion of social welfare so that a higher degree of inequality corresponds to a lower level of social welfare for a given total income. [Dalton (1920)] Normative: what should be. Positive: what is. Within normative approach inequality ceases to be an objective notion and measurement enmeshed with that of ethical considerations.
Setup Society has n individuals. Index i stands for a generic individual, i = 1, 2,..., n. Set of Social States X. Social Welfare Function W an ordering of the set of alternative social states.
Example of Social States Social states completely defined by amount of income received by each individual in society. Social State A = (y 1, y 2,..., y n ). Social State B = (z 1, z 2,..., z n ). To claim State A preferred to State B = W (A) > W (B).
Example continued 1 Let n = 3. 2 A = (20, 40, 60), B = (10, 20, 30) W (A) > W (B). 3 A = (10, 20, 30), B = (10, 10, 40) W (A)? W (B). W determines ranking of A and B in point (3).
Utilitarianism Jeremy Benthem (UCL) Greatest good for the greatest number. Pioneered approach; late eighteenth century. [ n W = max U i (x i ) ] x i = Y. x i i=1 i Utilitarianism sometimes thought of as an egalitarian social criteria. Problem set will investigate the properties of utilitarian social welfare function.
Focus on Descriptive Measures Compare relative inequality of two income distributions. Will highlight connection to Social Welfare Criterion.
Criteria The inequality measures are each a scalar x. But many scalar inequality measures exist and can and do give conflicting results. Consider a set of four criteria of desirable properties of inequality measures. Each criteria (weakly) relates to ethical judgement. Need to be cognizant of the relationship.
Anonymity principle It does not matter who earns the income. Anonymity because we care about the ordering but not the identity of each earner. All that matters is the ranking from lowest to highest: y 1, y 2,..., y n y (1), y (2),, y (n) order statistics y (1) y (2) y (3) y (n) Since identity of person doesn t matter we can dispense with the order statistic notation y (k) and use the simpler notation where index i represents the level of the i th income level in the society. y 1 y 2 y 3 y n
Example of Anonymity Principle Again n = 3. A = (10, 20, 30), B = (20, 10, 30) Anonymity: W (A) = W (B).
Population Principle If we compare an income distribution over n people and another population with 2n people with the same income pattern repeated twice, there should be no difference in inequality among the two income distributions. Anonymity states that no information is lost by retaining only the sequence of individual incomes (and not the identities of each) The population principle states it doesn t matter how large the population is, we can convert everything to percentiles (bottom 1%, lowest 20%, top 25%)
Example of Population Principle n = 3 A = (10, 20, 30) n = 6 B = (10, 20, 30, 10, 20, 30) Population Principle: W (A) = W (B).
Relative Income Principle Only the relative incomes should matter and the absolute levels of these incomes should not. Thus if we transform one distribution by multiply by a positive constant (e.g., Y 1 = λy 0 ) then inequality should be the same for the two distributions. Income levels have no meaning for inequality measurement. Absolute measure matters for assessing economic development. We will see that level matters for the measurement of poverty. Roughly think of poverty as a measure of location (level) and inequality as a measure of dispersion.
Parsimony Relative income principle means that data can be further collapsed. Population and income can be expressed as shares of the total. This means we can compare income distributions for countries with different average income levels. Need to record only income shares to measure inequality. Ordered incomes from poorest to riches. So if put into five income categories (quintiles) report the share of income by each fifth of the population. For example, Figure 6.3 (p. 177) graphs the share of income of the (richest) fifth quintile (31%), share of the fourth quintile (25%), share of the third quintile (20%), share of the second quintile (15%), and first quintile (poorest) (9%).
Dalton Principle The two principles not controversial, third more difficult (well being is proportional to income). This is fundamental to the construction of inequality measures. Let (y 1, y 2,..., y n ) be an income distribution and consider two incomes y i and y j with y i y j. A transfer of income from individual i to individual j is called a regressive transfer. If inequalities is strict y i < y j the regressive transfer is from the poorer individual to the richer individual. With weak inequality ( ) use the language not richer to not poorer
Dalton Principle Our inequality index as a function of the form: I = I (y 1, y 2,..., y n ) with I defined over all conceivable distributions of income (y 1, y 2,..., y n ). Dalton principle: if one income distribution can be achieved from another by constructing a sequence of regressive transfers, then the former distribution must be deemed more unequal than the latter. If for every income distribution (y 1, y 2,..., y n ) and every transfer δ > 0, I (y 1,..., y i,..., y j,..., y n ) < I (y 1,..., y i δ,..., y j + δ,..., y n )
The Lorenz Curve Lorenz curve is a simple diagrammatic way to depict the distribution of income. On the horizon axis we list the cumulative percentage of the population arranged in increasing order of income. Thus point on the axis refer to the poorest 20% of the population, the poorest half, etc. On the vertical axis we measure the percentage of the national income accruing to any particular fraction of the population thus arranged. The diagonal line (45 ) represents equal distribution income.
Lorenz Curve Properties The slope of the Lorenz curve is the contribution of the person at that point to the cumulative share of national income. Ordered from poorest to richest the marginal contribution can never fall. Equivalently, the Lorenz curve can never get flatter as we move from left to right.
Lorenz Curve The overall distance between the 45 and the Lorenz curve represents the amount of inequality present in the society.
Lorenz Curve Figure 6.4 1 0.9 0.8 0.7 0.6 Cumulative Income 0.5 0.4 0.3 0.2 0.1 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Cumulative Population
Lorenz Curve Comparisons 6.5 1 0.9 0.8 0.7 Cumulative Proportion of Income 0.6 0.5 0.4 L(1) 0.3 L(2) 0.2 0.1 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Cumulative Proportion of Population
Lorenz Curve: Crossing 1 0.9 0.8 0.7 0.6 Cumulative Income 0.5 0.4 0.3 0.2 0.1 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Cumulative Population
Complete Measures of Inequality Lorenz curves offer a visual representation of inequality. But the curves offer only a partial ranking: when Lorenz curves cross they offer no additional information to prefer one distribution over another. The (scalar) numerical measures provide a complete ranking of alternative distributions. However, these quantitative measures sometimes disagree.
Assume we have a population census. Scalar Measures of Inequality: Setup There are N units (households or individuals) in the population. Let there be m income groups, e.g., [10000, 20000) with representative value y g for group g. y g could be the mean income (within the interval), or the median, or the interval midpoint. Define the mean as µ = 1 N m g=1 Will divide by µ to make the inequality measures independent of the units of income. y g
Three simple measures of Inequality Range R = µ 1 (y G y 1 ), crude but sometimes useful. Kuznets Ratio Richest x% to the poorest y% where x and y stand for numbers such as 10, 20, or 40. Also crude, focuses on ends of distribution. Mean Absolute Deviation M = 1 Nµ m g=1 n g y g µ. Do not use.
Three Measures that Satisfy all Four Principles Coefficient of Variation C = m n g g=1 N ( ) yg µ 2. µ Gini Coefficient G = 1 2n 2 µ m j=1 m k=1 n jn k y j y k. Theil Index T = 1 Nµ m g=1 y g ln (y g /µ)
Comments on Measures CV This is a standard measure of relative dispersion. Gini It is the workhorse of inequality studies. The Gini Coefficient is the ratio of the area between the Lorenz curve and the 45 line to the area of the triangle below the 45 line. G = area between the Lorenz Curve and the 45 line area of triangle below 45 line
Graphical Display of Gini 1 0.9 0.8 0.7 B 0.6 Cumulative Income 0.5 A 0.4 0.3 0.2 0.1 Gini=A/B 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Cumulative Population
Comments on Measures Theil The Theil Index can be used to exactly decompose overall inequality into between group and within group components. Useful for descriptive studies to disaggregate inequality by race, gender, education, etc.